Upgrade managers for differential upgrade of distributed computing systems

Abstract

Examples of systems described herein may advantageously facilitate a software upgrade of one or more computing nodes of a distributed system without requiring a reboot of the node or otherwise rendering the node completely unavailable during upgrade. Upgrade portals described herein may provide each computing node with only the differential data needed to upgrade the node. Upgrade managers at each computing node may upgrade software at the computing node based on the differential data and restart services effected by the upgrade using the differential data. Other services may remain available during the restart of the effected services.

What is claimed is: 
     
         1 . A distributed system comprising:
 a first upgrade manager hosted by a first computing node of a distributed computing system, the first upgrade manager configured to:
 receive first differential upgrade data for a first selected package of a software upgrade based on a comparison between existing packages hosted by the first computing node and packages of the software upgrade, and 
 upgrade the first selected package; and 
   a second upgrade manager hosted by a second computing node of the distributed computing system, the second upgrade manager configured to:
 receive second differential upgrade data for a second selected package of the software upgrade based on a comparison between existing packages hosted by the second computing node and the packages of the software upgrade, wherein the second differential upgrade data is different from the first differential upgrade data, and 
 upgrade the second selected package. 
   
     
     
         2 . The distributed system of  claim 1 , wherein the first upgrade manager is further configured to restart a first service of a plurality of services, the first service utilizing the first differential upgrade data, and wherein the second upgrade manager is further configured to restart a second service of the plurality of services, the second service utilizing the second differential upgrade data. 
     
     
         3 . The distributed system of  claim 1 , further comprising an upgrade portal configured to compare the existing packages hosted by the first computing node and the packages of the software upgrade and provide the first differential upgrade data. 
     
     
         4 . The distributed system of  claim 3 , wherein the upgrade portal is hosted by the first computing node or the second computing node. 
     
     
         5 . The distributed system of  claim 3 , wherein the upgrade portal is hosted by a computing system other than the first computing node or the second computing node. 
     
     
         6 . The distributed system of  claim 3  further comprising an upgrade configuration file associated with the packages of the software upgrade and a first configuration file associated with the existing packages hosted by the first computing node, and wherein the upgrade portal is configured to compare the upgrade configuration file and the first configuration file to provide the first differential upgrade data. 
     
     
         7 . The distributed system of  claim 6 , wherein the upgrade portal is configured to compare a checksum of the upgrade configuration file with a checksum of the first configuration file to provide the first differential upgrade data. 
     
     
         8 . The distributed system of  claim 2 , wherein other services of the plurality of services remain available during restart of the first selected service of the plurality of services. 
     
     
         9 . A method comprising:
 comparing packages of upgraded software to packages hosted on respective computing nodes of multiple computing nodes in a distributed system to generate first differential upgrade data for a first one of the multiple computing nodes comprising a first set of selected packages of an upgraded software and to generate second differential upgrade data for a second one of the multiple computing nodes comprising a second set of selected packages of the upgraded software, wherein the second differential upgrade data is different than the first differential upgrade data;   upgrading the first set of selected packages on the first computing node; and   upgrading the second set of selected packages on the second computing node.   
     
     
         10 . The method of  claim 9  further comprising restarting selected services of the multiple computing nodes impacted by the first differential upgrade data and the second differential upgrade data while maintaining availability of other services of the multiple computing nodes. 
     
     
         11 . The method of  claim 9 , wherein the first differential upgrade data contains less data than the upgraded software. 
     
     
         12 . The method of  claim 9 , wherein comparing packages comprises comparing a configuration file of the upgraded software with a configuration file associated with currently-installed packages hosted on a respective one of the first and second computing nodes. 
     
     
         13 . The method of  claim 12 , wherein comparing packages comprises comparing a checksum of the configuration file of the upgraded software with a checksum of the configuration file associated with the currently-installed packages hosted on the respective one of the first and second computing nodes to provide the first and second differential upgrade data, respectively. 
     
     
         14 . The method of  claim 12  further comprising transmitting the first differential upgrade data from one of the multiple computing nodes to others of the multiple computing nodes. 
     
     
         15 . The method of  claim 12  further comprising upgrading software at the multiple computing nodes using a respective differential upgrade data. 
     
     
         16 . At least one non-transitory computer readable media encoded with instructions which, when executed cause a computing system to perform operations comprising:
 receiving, at an upgrade portal, first data indicative of installed software packages currently hosted by a first computing node of a distributed system and second data indicative of installed software packages currently hosted by a second computing node of the distributed system; and   providing, from the upgrade portal, first differential upgrade data comprising first selected software packages for upgrade, the first selected software packages selected based on a comparison of the first data with packages of a software upgrade; and   providing, from the upgrade portal, second differential upgrade data comprising second selected software packages for upgrade, the second selected software packages selected based on a comparison of the second data with packages of the software upgrade, wherein the second differential data is different than the first differential upgrade data.   
     
     
         17 . The at least one non-transitory computer readable media of  claim 16 , wherein the operations further comprise causing the computing system to cause a restart of selected services hosted by the first computing node based on the first data while maintaining availability of other services during restart of the selected services. 
     
     
         18 . The at least one non-transitory computer readable media of  claim 17 , wherein receiving the first data comprises receiving a configuration file indicative of the installed software packages currently hosted by the first computing node. 
     
     
         19 . The at least one non-transitory computer readable media of  claim 18 , wherein the first differential upgrade data is based on a comparison of a checksum of the configuration file and a checksum of another configuration file associated with the packages of the software upgrade software upgrade. 
     
     
         20 . The at least one non-transitory computer readable media of  claim 16 , wherein the upgrade portal is hosted by one computing node of the distributed system. 
     
     
         21 . The at least one non-transitory computer readable media of  claim 16 , wherein the operations further comprise upgrading software of each of the first and second computing nodes of the distributed system using a respective differential upgrade data of the first differential upgrade data and the second differential upgrade data.